糖心TV

Sara Wade, Department of Statistics, University of 糖心TV

Abstract:

In this talk, I will discuss Bayesian nonparametric modelling in machine learning, with particular focus on Gaussian process models. A review of Gaussian process models for supervised learning tasks will be provided, along with a discussion on its connections to classical and parametric models. I will then introduce the Gaussian process latent variable model for unsupervised learning tasks and discuss its connections with probabilistic principal components. Finally, I will describe my research on a fully Bayesian inference algorithm for a supervised version of the Gaussian process latent variable model.